Fourth International Conference on Plant Protection in the Tropics
PROCEEDINGS
PLANT PROTECTION
IN
THE TROPICS
Editors:
A. Rajan
Yusof Ibrahim
Systems approaches in pest management : The role of production ecology
R. Rahbinge, WA.H. Rossing and W. VanderWerf . • . . • . • . • • . . . . . • . .
Published by:
Malay~!an
Plant Protection Society
1994
25-46
4th International Conference on Plant Protection in the Tropics
Plenary paper
28-31 March 1994, Kuala Lumpur, Malaysf.a.
Systems approaches in pest management:
l,he role of production ecology
R. RABBINGE, W.A.H. ROSSING and W. VANderWERF
Departme11t of Tlteoretical Production Ecology,
lrVageni11gen Agricultural University, PO Box 430, 6700 AK Wagenittgen, The Netherlands.
INTRODUCTION
Systems research and simulation models
Production ecology focuses on increasing
insight in biological production processes at
different hierarchical integration levels and
expressing this insight in quantitative models.
Relevant integration levels in production ecology
are indicated in Figure 1. Each integration level
is characterised by its temporal and spatial
dimensions. In the production ecological systems
approach, the central idea is that one must
identify, delineate ru1d understru1d a system in
order to be able to influence it in a predictable
manner (Spedding, 1990). Models are used as
bridges between integration levels. They allow a
better grasp of the behaviour of the system by
using qmmtitative descriptions of processes or
input-output relations at lower integration levels.
In this way, knowledge at lower integration
levels is scaled up to higher levels.
Mission statement of production ecology
The population of the world in 1994
amounts to approximately 5,400 million people,
and increases annually by about J 00 million.
Projections by the UN for the year 2050 show a
population size between 8,000 and 12,000
million. Stabilisation of the population around
11,000 million is anticipated. The increase in the
size of the population io combination with the
demand for a more 'luxurious' food package,
require growth of agricultural production. At the
same time locally occurring over-exploitation of
natural resources must be avoided and, where
necessru·y, reversed. Hence, agriculture needs to
meet a rising demand for marketable output,
while satisfying ever tighter constraints with
respect to toxicological safety of the product and
impact of production techniques on nature,
Time (years)
environment ru1d landscape. These issues require
a scientific analysis of socially relevant options
for agricullural production activities at different
levels of integration. Production ecology is an
integrating scientific discipline which aims at
exploring options for sustainable primary and
secondary production systems (Rabbinge, 1986).
of
these
Insight
in
the
functioning
agro-ecosystems is based on quantitative
understanding of physical, chemical and
eco-physiological processes that take place in
such
systems.
Various
ecological
and
socio-economic objectives and constraints
10"8
10"6
10-1
10"2
10°
10 2
104
10 6
dm 2
m2
a
ha
km2
Area (km 2)
determine the possibilities of these systems.
Sustainability is used here in the sense that the
possibilities for utilisation of the natural
Figure 1.
Time and spatial scales of
---res0urees--soil-,-water andairarenot deteriorated -- integration - levels; relevant - in- -production
ineversibly by current human activities.
ecology.
25
. A simulation model is the product of an
analytical phase, in which a system is
decomposed into elements of which the
behaviour is studied, and an integrative phase, in
which models are used to synthesise the acquired
knowledge. The analytical process leads to
discoveries, new hypotheses and new insights at
every lower level of integration. The
synthesis-oriented approach leads into lllC
opposite direction and results in understanding at
every higher level of integration. The level of
detail required in desctibing the underlying
processes is determined by the objectives at the
higher integration levels. Exploration of options
for sustainable production requires thorough
knowledge of elementary processes, useful
models for integration of process knowledge and
a continuous interaction between model and
reality on experimental farms and in practice.
Therefore, the analytical and the synthesising
scientific approach in production ecology arc
complementary (Rabbinge and de Wit, 1989).
The systems approach is juxtaposed to the
statistical approach in which a pat1icular aspect
of system behaviour is coiTelated to other aspects
at the system level. The statistical approach
results in descriptive models, which lack
inherent causality.
Crop protection at different integration levels
Crop protection issues can be studied at
different integration levels: pathosystem, crop
system, cropping system, fat·m system, region
a~1d supra-regional level. Much disciplinary
entomological and phytopathological resemch is
conducted at the lower integration levels of cells,
subcellular structures
and
biomolecules.
Processes at adjacent integration levels have
time coefficients (the reciprocal of the relative
rates of change) that typically differ by one or
two orders of magnitude, i.e. a factor 10 to 100.
Simulation models at·e used to bridge two or
maximally three levels of integration. Usually,
lack of knowledge and Iat·ge differences in
temporal and spatial scales prohibit bridging of
more than three integration levels. Moreover, the
explanatory power and usefulness of a model
may not increase substantially, or even decrease
-when--there--comes -a -large--gap -oet ween-the
explanatory and explained level. lllUs, plant-host
interactions at the molecular level are seldom
represented in epidemiological and crop growth
Problems
of aggregation
are
models.
encountered when research results at the crop
level need to be scaled up to the cropping system
level. For example, at the cropping system level
information is needed on population dynamics of
the weed seed bank, which is of little concern for
operational weed management during the
growing season and has received little research
attention for that reason.
The division of pest mat1agement issues
into policy-oriented, strategic, tactical and
operational (Conway, 1984) is based on
differences in time and spatial scales.
Policy-oriented issues concern the farm and
higher integration levels, with time coefficients
of at least several years. Examples are the
governmental policy statements on crop
protection in various countries, initially
especially in South and South-East Asia. A
recent culmination point is the decimation of
'Agenda 21' at the 'Earth Summit' in Rio de
Janeiro, in which integrated pest management is
embraced implicitly or explicitly (Table 2 in
Zadoks, 1993). Strategic, tactical and operational
questions may apply to the same spatial scale,
e.g. a farmer's field, but they differ in temporal
scale. Strategic questions at the field level
address issues covering several growing seasons,
while tactical questions concern effects within
one growing season. Operational questions
address 'how-to' problems and concem day to
day implementation of decisions made at the
tactical and strategic levels.
Irrespective of the integration level, crop
protection issues comprise four elements,
represented in the disease tetrahedron (Zadoks
and Schein, 1979; Figure 2): the growth reducing
factor or pest, the crop, weather or climate, and
the human actor, be it a fatmer or a policy
maker. 1l1e pest interacts with the crop. The
outcome of the interaction is determined by pest
and crop attributes, which are influenced by
weather and the human actor. Studies of such
complex di- and tri-trophic systems easily lead to
situations where 'everything affects everything'
at ll1e system level so that causes and effects
-cannor-_re_ ·ursennmgieu~-Fforrc-a--Jrroou:ct1on
ecological
perspective,
it
is
essential
to
A: Population dynantics
climate
weather
,.···'
;:.'
growth
reducing factor~----------
crop
B: Injury and dantagc
climate
weather
'•,',,,
'~
growth
reducing factor
__________ ..,_
crop
crop growth models, which explain crop growth
and production and the way these are affected by
injury (Figure 2b). Such studies of population
dynamics and damage provide the building
blocks for pest management systems in which
different slmtegies of human intervention are
analysed with respect to the objectives of
management (Figure 2c ).
In this contribution we will demonstrate the
need for a production ecological approach in
crop protection. First, crop ecological principles
of primary production will be discussed. Next,
attention will be focused on management of
agricultural production systems and the role
production ecology can play in improving
systems management. Finally, we will discuss
institutionalisation of the production ecological
approach in crop protection research in Asia
tlu-ough a training and research project in rice
production systems.
PRINCIPLES OF PRIMARY PRODUCTION
C: Managetnent
Production situations, production
technologies and production levels
,'~',
In the analysis of primary (and also
,'' : ',,
secondary) production systems, distinction is
,''
:
',,
,•'
,·
'•,
made between production situations, production
,'
',
technologies and production levels (Rabbinge,
.. /
climate
',.,
1993).
;:/ /weather ~<.
The production situation at a specific site is
growth
,..
..
characterised
by physical factors: climate, water
reducing factor ....
..crop
availability and aspects of the soil such as
compaction, stoniness, water holding capacity
and steepness. Together, these factors determine
Figure 2.
The disease triangle with emphasis
the conditions for agricultural production. Soils
on (A) population dynamics of the growth
characterised by high water holding capacity,
reducing factor, (B) injury and damage, and (C)
high fertility and low stoniness in combination
crop protection.
with high water availability allow higher yields
than soils which have low water holding
capacity, low fertility and high acidity or salinity
distinguish population dynamics of the pest, crop
damage, and management of the crop-pest
in combination with poor water availability.
Production technology is the total of production
system. Weather, nutritional state of the crop,
and human interventions are input for models
techniques and methods. This encompasses all
which explain temporal and spatial population
cropping measures from before sowing till after
dynamics (Figure 2a). In studies of damage, on
harvest, such as ploughing, seed bed preparation,
-th€-Gther-hand,wcathcr-,----population-dynamics-of- s_2wing, weed control, etc. The combination of
the pest, and human interventions are input for
production situation and production technology
farmer
I
I
'
27
--------
_-----------
-----------------------
results in a production level. The production
level represents the amount of product per
hectare. In good production situations, a high
production level can be realised with a given
production technology. In poor production
situations greater inputs are needed to reach the
same production level, if possible at all. As a
consequence, production in poor production
situations is less efficient, both per unit area and
per unit product than production in good
production situations.
The production level is influenced by three
categories of growth factors (Figure 3)
GrowU1 defining factors
These determine the potential yield which
is realised when crops grow with an ample
supply of water and nutrients. Growth defining
factors include site-specific environmental
variables, such as temperature and incident
radiation which depend on latitude and day of
the year, and species-specific characteristics
concerning physiology and geometry of the
leaves and the roots, and phenology. Potential
growth rates appear to be 2-3 J..lg (dry matter) J-l
(light) when expressed per unit of light, by
definition the only limiting resource under
optimal conditions. In temperate regions,
potential growth rates range from 15 to 35 g dry
matter per square meter per day. Situations
where potential growth rates are reached are
rare, and may occur only in protected cultivation.
Growth UmiJing factors
These comprise abiotic resources such as
water and nutrients, which limit the growth rate
of the crop to a value below the maximum when
their supply is sub-optimal. TI1e associated yield
level is called 'attainable'. In conjunction with
C02 uptake, water vapour is released through the
stomata. The rate of transpiration depends on
incoming radiation, vapour pressure deficit of
the air, and stomatal aperture. About 150 to
300 g of water is transpired for each gram of dry
matter
assimilated.
In
other
words,
approximately 4 to 8 mm water are required per
day to maintain a potential growth rate of 25 g
(dry matter) m-2 (ground) day-1. The nitrogen
requirement for potential growth is in the order
of 10 g m-2 (ground), assuming an LAI of 4m2
(leaf) m-2 (ground), a specific leaf area of 20m2
(leal) kg-1 (leaf dry matter) and a leaf nitrogen
production
level
Defining factors
potential
A
potential
~
.!!~t~!~.~~!~............... .
actual
attainable
f.--------
-l
Limiting factors
radiation
temperature
crop phenology
physiological properties
crop architecture
water
nitrogen
phosphorus
pests
diseases
Reducing factors weeds
pollutants
calamities
actual
1----------
bad production situation
Figure 3.
good production situation
Production situation, production levels and associated principal growth factors.
content, necessary for maximum photosynthesis
rates, of 5% on a dry matter basis. These values
are only rules of thumb, and vary according to
latitude, crop species and other factors. Van
Keulen and Wolf ( 1986) present more accurate
methods for estimating water- or nutrient limited
growth rates.
Growth reducing factors
Unfortunately,
crop
ecology
is
an
underdeveloped aspect in training programs for
crop protection students throughout the world.
Biotic growth reducing factors: Mechanisms
of damage
The simplified example in the previous
section shows that yield loss due to insects,
pathogens and other growth reducing factors is a
complex function of a large number of variables
and depends on the production level.
Experimental studies alone are inadequate to
unravel the way in which complex systems with
many interacting elements behave. Fortunately,
explanatory simulation models provide powerful
tools for identifying damage mechanisms and
quantifying
their relative and
absolute
contributions to yield loss.
Pathogens, insects, weeds, pollutants of air,
water and soil, and extreme weather condHions
represent growth reducing factors. Due to these
factors crop production is reduced from the
attainable yield level to the 'actual' yield level.
Especially with regard to pathogens, insects and
weeds a wealth of information has been collected
during the last 15 years concerning the way in
which yield reduction is caused. Damage, i.e. the
size of yield reduction, depends on the plant
processes which are affected, i.e. the damage
mechanisms, the growth rate of the healthy crop
and the timing and intensity of growth reduction.
growth rate
Thus, damage by biotic growth reducing factors
radiation
is the result of crop physiological, crop
ecological and population dynamical processes.
Experimental results in the Sahel (West
high
Africa; Figure 4) illustrate the interdependency
water
of the effects of growth defining and growth
limiting factors on crop growth. In the period
June- July, directly after germination, water sets
a maximum to the growth rate. From July till
August, the low availability of phosphorus is the
most limiting factor, while towards the end of
the growing season the availability of nitrogen is
the most limiting factor. Growth ends when the
low
crop reaches physiological maturity, which is
mainly a function of temperature. Because
several limitations act simultaneously, human
j11n feb mar apr may jun
jul aug sep oct nov dec
intervention aimed at alleviating the limitation
months of year
by any one of the limiting factors would produce
only modest increases in crop production. Pest
•""'igure 4.
The effect of the relative
management always operates within the
availability of four principal growth factors constraints set by the growth defining and
radiation, water, nitrogen and phosphorus - on
growth limiting factors. Damage due to, for
plant growth in the Sahel. At any moment in
example, early senescence caused by a disease
time the uppermost line represents the potential
will depend on the production level at which the
growth rate, while the bottommost line
crop is growing. Hence, crop protectionists need
to have quantitative insight in crop growth
represer~ts ~1e actual growth rate. The actual
_._ ___ . __
_ _ _ p_r_od~_l!PJLIS_equal_tQ_llle shaded____W"ea. Aftet
_
___ __ __ _ ___ __ ._
processes as well as eptdemmlogrcal processes,
p
. d V.
dV K
( - ) ---.
.
.
ennmg e nes an
an eu 1en 198 2 .
and the1r respective relations to external factors.
29-·
Crop growth at the potential production
level can be modelled at different levels of
physiological detail (Spitters, 1990). A summary
approach is based on empirical evidence that in
many crops growing at the potential production
level, the rate of growth is approximately
proportional to the amount of light intercepted
by green foliage (Monteith, 1977). A plot of total
biomass against cumulative intercepted light (LI)
results in a straight line. The slope of this line
represents the average crop light-use efficiency
(CLUE). Growth reducing factors may reduce
Ll, CLUE or both. When CLUE is unaffected by
injury, the photosynthetic capacity of the green
area is apparently unaltered. Injury then simply
causes a reduction of the amount of energy
available for crop growth and damage is
proportional to the amount of energy foregone.
When CLUE is decreased, the relation between
damage and intercepted light is more complex.
This warrants analysis of damage with a more
comprehensive ceo-physiological crop growth
model.
In a comprehensive model of crop growth
and development (e.g. SUCROS, SpiUcrs el a/.,
1989) daily rates of gross photosynthesis are
simulated by combining the light profile in the
crop with the light response of carbon
assimilation by individual leaves. Gross carlxm
assimilation is conveniently described by a
saturation curve, which is characterised by an
initial light usc efficiency and an asymptote at
high light intensities (Figure 5). The rate of daily
dry matter production is found by subtracting the
rate of maintenance respiration from the
calculated gross photosynthesis rate and
accounting for the costs of conversion of
assimilates into structural dry matter of the
various organs. Integration over the days in the
growing season results in total dry matter
production. Damage mechanisms can be
introduced into SUCROS at the level of whole
plant processes (Figure 6).
Analysis of the effect of a growth reducing
factor on crop growth and production proceeds
along the steps shown in Figure 7. First, an
inventory is made of the crop growth processes
-thaLare-affected-by-the_gro_wtiLrcducing_factor
(Rabbinge and Rijsdijk, 1981). Next, these
(likely) -damage mechanisms are ranked
according to their relative importance. The most
important ones are quantified using literature
information and experiments. These damage
mechanisms are then introduced into a crop
growth model. Evaluation of the performance of
the model shows to which extent the major
causes of damage are understood, and whether
additional damage mechanisms need be
considered. When the model is in sufficient
agreement with reality, it may be applied to
situations for which no or limited experimental
information exists. Such a model may, for
instance, be used to study the effect of timing of
disease on yield loss under different weather
conditions. The level of detail in the crop model
is dictated by the objectives of the study and the
characteristics of the pathosystem. For instance,
it may be necessary to distinguish different leaf
layers to account for the vertical distribution of a
pathogen. Technical details of the approach are
described in Rabbinge el al. (1989).
Rate of photosynthesis (mg CO 2 m •2 s •1 )
1.2 ..-----.----.------.---.--.......---..---.
-0.2 L . -_
0
__.__
__.L_
100
_ _ _ . . _ __
_.1..__
200
_ . __
___J
300
Absorbed light (J m -2s -1 )
Figure 5.
Illustrdtion of a photosynthesis
versus light response curve, characterised by the
parameters Pnl' e and Rd. The curve is described
by a negative exponential equation.
A typology of damage mechanisms can be
drawn UQ which reflects the nature of th(!
interaction between growth reducing factor and
crop growth processes:
30
I
0
1
I
I
!
Ii
maintenance
respiration
~
Q)
growth
respiration
growth rate
..............................................
Figure 6.
Schematic representation of an explanatory approach to modelling potential crop growth
and injury, utilising the light profile within the canopy, photosynthesis characteristics of individual
leaves, respiration and dry-matter partitioning. Arrows indicate potential damage mechanisms: (1)
competition for light, (2) decrease of the rate of photosynthesis, (3) effects on respiration, (4) plant
~-~~-~-~-------~-~~~-~~~~~~~~~ _____killing,_(i)_tissllC necrosis and interception of lighb_({j} tissue consumption, (7) assimilate
consumption, (8) hampering of water uptake, (9) induction of hormonal effects on stomatal regulation,
and{l0)deformations;
statistical approach based on the density-damage
relationship of one year would not have
predicted events in the other year.
Meld biomass
•. ol weed- free
·-----
--:-----____.
Validation of
model
performance
0~--------~,0~0--------~--------~
200
300
Echlnochloa plan I 1 m-2
Figure 8.
Final above-ground biomass of
maize in 1982 (•) and 1983 (•), expressed as
percentage of weed-free control, in dependence
of initial density of E. crus-galli (Kropff et a/.,
1984).
proceed to
applied
research
Figure 7.
Schematiscd
procedure
for
assessment of the effect of damage mechanisms
(dm) on crop growth and production (Bastiaans
and Kropff, 1991).
Decrease of the rate of photosynthesis
Many leaf pathogens and viruses decrease
the photosynthetic capacity of the leaves they
attack. Mites which cause minute injuries by
sucking up epidermal cell contents, have a
similar effect. The effect of different growth
reducing factors on leaf photosynthesis can be
described by a simple descriptive equation with
one parameter (Bastiaans, 1991). The parameter
in the equation, (3, represents the ratio of a (real
or imaginary) influence area of a lesion and its
visible surface. The equation is valid when
lesions are randomly distributed over the leaf.
Application of this model to a number of growth
reducing factors showed that for some leaf
pathogens, such as Phytophthora infestans (Van
Oijen, 1991), Septoria nodorum and Puccinia
recondita, and for avocado brown mite,
0/igonychus punicae (VanderWerf eta/., 1990)
the decrease in leaf photosynthesis is
proportional to the area covered with lesions, i.e.
Competition
Weeds compete with crops for light,
nutrients and water. The amount of damage
depends strongly OQ the timing of attack. An
example is found in field experiments by Kropff
et a/. (1984). Maize biomass was lower at high
densities of Echinoch/oa crus-galli than at low
densities. However, while a density of 100
E. crus-ga/li plants per m2 caused a yield loss of
8% in 1982, it caused a yield reduction of 88%
in 1983 (Figure 8). Analysis with a
comprehensive simulation model for crop-weed
interaction showed that the difference between
the two years was explained by year to year
variation in time of emergence of weed and
maize plants: in 1982 the crop emerged 5 days
earlier than the weed and did not suffer from
-competition-fCJr-light.-In-l-98J,-w~~dS-~m~..g~d-at~-thek_f3_...wlucs_are___no1___S_ignifkantly_dif[erent
the same timeJ\S the crop, which resulted in
from 1. Similarly,forbeetyellowing viruses the
intensecompetition- for ·Jight (Spittcrs~--1984). A
fractional decrease in crop ·photosynthesis---is
32
present experimental research is carried out to
elucidate these relations further (Rossing et al.,
1993).
approximately equal to the fraction incident light
which is intercepted by yellow leaves. For other
growth reducing factors, for example powdery
mildew, Erisyphe graminis, in wheat effects on
photosynthesis are much larger than the fraction
leaf area covered with lesions. l11cse effects of
mildew are thought to be caused by hormonal
effects on stomatal regulation, since the initial
light use efficiency of leaf photosynthesis is
proportional with lesion cover, i.e. p equals 1,
whereas for the maximum rate of photosynthesis
the P-value is 6. In contrast to light utilisation at
low light intensities, the maximum rate of
photosynthesis is limited by the C0 2 supply and
therefore by stomatal aperture. For leaf blast in
rice, Pyricularia oryzae, P-values for both the
maximum rate of photosynthesis and the initial
light use efficiency exceed I (Bastiaans, 1991).
Also for peanut leafspot (Cercospora sp.)
P-values for leaf photosynthesis appear to exceed
1 (VanderWerf eta/., 1991).
Tissue necrosis and interception of light.
Most leaf pathogens and some viruses
cause local lesions. Eventually these lesions die
off. Damage usually exceeds the loss of biomass
because the dead spots continue to intercept
light, and therefore, compete with other,
productive tissues for this scarce resource. As a
result the vertical position of the dead tissue in
the canopy determines to a large extent the
amount of damage. Van Roermund and Spitters
( 1990) analysed an experiment on yield
reduction among seven winter wheat genotypes
due to brown rust, Puccinia recondita, using a
mechanistic crop growth model. Depending on
the genotype, inoculation at pseudo-stem
elongation resulted in epidemics of different
intensity, the area under the curve ranging
between 0.1 x lo4 to 19 x J04 pustule-days
culm-1. Accelerated leaf and ear senescence due
to brown rust accounted for 91% of total ·
damage. Light interception due to leaf coverage
by pustules accounted for less than 1%. The
remainder of damage is due to assimilate uptake
by the fungus (see below). Such analysis
represents 'experiment-added-value' by yielding
insights in the relative importance of damage
mechanisms that could not have been obtained
experimentally.
Effects 011 respiratio11.
For many microscopic pathogens which
multiply in leaves, an increase in leaf respiration
has been described. It is often unresolved Lo
which extent the increase is caused by processes
related directly to the growth and reproduction of
the pathogen, or by defense mechanisms of the
plant.
Pla11t killing
Some pathogens, for example root fungi
during seedling stages, kill entire plants. Such
diseases cause substantial damage unless the
crop is capable of compensating by increased
tillering so that complete light interception is
attained with a lower plant density. Killing of
entire plants at later development stages causes
damage which is approximately proportional to
the numhcr of plants lost. Using a well-tested
comprehensive crop growth model, Rubia and
Penning de Vries ( 1990) showed that killing of
up to 20% of the tiller of rice cultivar IR64 at the
vegetative stage, as may occur due to stem borer
attack, had no important effect on grain yield,
while yield loss almost equalled the proportional
loss of tillers when injury occurred at the grain
filling stage. Compensatory ability depends on
Tissue consumption
Many growth reducing factors consume
plant organs, especially leaves. The amount of
damage depends on
the compensatory
possibilities, which are a function of timing of
attack, type of organ consumed and state of the
crop. When the leaf area index is high, leaf
consumption hardly reduces crop light
interception and, photosynthesis and growth rate.
At excessively high leaf area index values, leaf
consumption can theoretically even increase
growth, due to reduced maintenance respiration
as a result of the reduction in biomass. Leaf
consumption in the initial phase of exponential
increase in leaf area can be extremely damaging
because the resulting decrease in the relative
~-~~-~~--~--=:-::-:-~~~---=~~__.._--"--~----:--o-:-:--------:~-----=:-----~~---::--~~-----:-~~----c::-~------=~~~~----"'c-~~~
==cum:Y=ru\~=¥-ieia=le'\lcCruid=-Jirm!~.oL=attacl<.:At·---=
33
-tliemorrrencof
grtiwm=t~ate=teads-=m=a=aelay1n•_
full light interception. As a result, potential
growth rates are realised during a shorter period
of time. Consumption of storage organs is very
damaging,
since
plants
usually
lack
compensatory potential. An indirect effect of leaf
consumption is that the competitive position
with respect to weeds deteriorates.
Assimilate consumption
Aphids suck up phloem sap and decrease
the amount of assimilates available for
respiration and growth. In addition, uptake of
amino compounds may disrupt nutrient balances.
During the exponential growth phase, assimilate
tapping insects can delay the growth of leaf
canopy as was shown in experiments (Hurej and
Van der Werf, 1993) and through modelling
(Groenendijk et a/., 1990). Assimilate lapping
during the linear growth phase decreases the
daily growth rate, especially due to uptake of
sugars. A large part of the sugars consumed by
aphids is excreted in the form of honeydew,
because aphids regulate their rate of uptake of
phloem sap such that they obtain sufficient
amounts of amino acids, which occur only in low
concentrations in phloem sap. The excreted
honeydew seals off stomata, it can accelerate
ageing of leaves and it enhances the
pathogenicity of saprophytic and pathogenic
fungi. Combined, these mechanisms can reduce
the rate of photosynthesis. Detailed production
ecological analysis showed that honeydew
excreted by Sitobion avenae in winter wheat
constitutes an important damage component,
especially at very high production levels
(Rossing, 1991).
Bastiaans and Teng (1994) showed that
assimilate uptake by sporulating lesions of the
fungus Pyricularia oryzae in rice under certain
conditions accounts for about 20% of the
damage. Assimilate uptake by P. recondita in
winter wheat was found to account for 9% of
total damage caused by brown rust (Van
Roermund and Spitters, 1990; see above).
stomatal closure and reduced crop transpiration.
Concomitantly, assimilation rate is reduced.
Induction of hormonal effects on stomatal
regulation.
Cyst nematodes appear to change stomatal
behaviour of injured plants, even before water
uptake is substantially decreased. The hypothesis
has been put forward that release of abscissic
acid by the roots in response to nematode
colonisation induces these stomatal effects
(Schans, 1992).
Deformations
Sucking insects often cause deformations.
Sometimes, components with hormonal activity
in the saliva of t11e insects play a role. Organ
deformation itself can be a damage component if
it concerns the harvestable parts. Leaf
deformations such as aphid-induced curling can
cause damage indirectly by decreasing light
interception.
The typology of damage mechanisms is not
exhaustive and may change as knowledge
accumulates about the ways in which biotic and
abiotic factors reduce crop growth. The
examples
illustrate
the
major
damage
mechanisms and underscore the importance of a
systems approach in which models are used to
increase understanding of the relative importance
of various interacting processes for damage.
Those same models are practically valuable
because tl1ey may be used to explore how
damage is affected by environmental conditions,
the timing of attack and by control measures at
different times.
Biotic growth reducing factors: Population
dynamics
The size of yield reduction caused by biotic
growth reducing factors is a result of their effect
on crop growth processes and their population
dynamics. Broadly, in population dynamical
analyses emphasis may be on phenology, the
development of individuals through various life
stages, or on intensity, the time course of the
density of the growth reducing factor in time and
Hampering of water uptake
Root pathogens, such as Verticillium spp.,
-------·-------and--alse--m:~matoo~s,--r~duG~-r..()p-growth-by--------Space.-Undcr,~t.anding-of-the-Jlrocesses-whim .._______________
_affectingtheuptake
dctcrmiue dcvelQPillQilt js .st.ill
oLwa1:er.Thisniayresultin
34
Jll{iillieJ1tWY·
··--·-..
However, for many growth reducing factors
adequate models have been developed which
describe the relation between phenology and
external factors. Usually, temperature is the
dominant factor governing developmental
processes once a growth reducing factor has
established a feeding relationship with the host.
To allow infection, special environmental
conditions may be required, for example with
regard to relative humidity, leaf wetness period
and photoperiod. In many crop systems
descriptive phenological models have found a
place in predicting the timing of critical events
for crop protection at the field level. A recent
overview of approaches and applications in
western Europe is given by Secher eta/. (1993).
The current understanding of the dynamics
of the intensity of many growth reducing factors
is still poor, mainly due to lack of insight in the
biotic mortality factors that operate at the field
level. From the progress made in entomological
research on biological control it is evident that
even in systems involving only one pest and one
predator or parasitoid a large number of
variables determines pest mortality. Simple
approaches such as the classical Lotka- Volterra
models arc of some value in exploring general
properties of biological control systems (Janssen
and Sabelis, 1992; Van der Werf et a/., 1994),
but they fall short in analysing the behaviour of
specific systems because they disregard
environmental influences, such as temperature,
and important behavioural components at the
individual level. Mechanistic models, that
include descriptions of behaviour at the
individual level arc better suited fo. analysing
behaviour of specific systems. The behaviour of
predators and parasites is governed by their
'motivational state', which is strongly related to
and can be modelled as the contents of a
predator's gut (Fransz, 1974; Rabbingc, 1976;
Sabclis, 1981; Mols, 1993) and a parasite's
ovariolcs (Van Rocrmund ru1d Van Lenteren,
1994 ). Pest mortality due to predation or
parasitisation is simulated depending on abiotic
environmental conditions and pest distribution.
Sensitivity analyses show how the likelihood of
----·-·---LhLL'iclLJllULogicaLcnntrol is_affcctcd by LmjJs_oLthe
.J?lqtlth()~(,pg~t.or natural cn~I!lY. Sitnulations
the insect baculovirus SeNPV in a population of
beet army worm, Spodoptera exigua, in
chrysanthemums show at which time virus
sprays can be best applied to obtain highest pest
mortality (Van dcr Werf et a/., 1991). Also for
pathogens, mechanistic approaches lead to better
understanding of system behaviour. Van der
Wcrf modelled spatial spread of virus yellows in
sugarbcct by simulating walking behaviour UIJd
virus transmission of individual Myzus persicae,
the vector (Van der Werf et a/., 1989). The
results of the preliminary model show a striking
asynchrony between the presence of aphids and
the appearance of yellowing symptoms (Figure
9), ru1d indicate that there is scope for reducing
the rate of virus spread by biological control of
the aphid vector. In another example of a
production ecological study Van Oijen (1991)
set out to identify the major characteristics of
potato cultivars which affect yield reduction by
Phytophthora bifestans. Sensitivity analyses
showed that relative growth rate of a
Phytophtora epidemic is determined to a large
extent by the radial growth rate of the lesions.
Since radial lesion growth is a process for which
genotypic variation exists in Solanum. spp., such
production ecological studies are now used to
guide breeding progmmmcs.
So far, only approaches ru1d applications at
the field level were discussed. At the regional
and supra-regional levels equally promising
advances have been made. Combination of
phenological models and climatic data using
Geographical Information Systems provide
insight into the risk of establishment of exotic
pests (Sutherst, 1991). Such quantitative
approaches can be used to focus quru·antine
efforts. When used in combination with different
climatic change scenario's, changes in risk can
be explored. An example is the assessment of the
risk of re-establishment of malaria and malaria
vectors in southern Europe, as a result of
anticipated global warming (Jclten ru1d Takken,
1994 ).
Understanding of epidemics at fieldregional and supra-regional scale has benefited
greatly
from
relatively
simple
and
mathcmaticallx tractable models for the rate of
~dy~t1C~<Jf . tl1eepi~en1ic .wave fronts . ru1d the
--with-a-mcchanistie-rnoclel-of-Hlc-cpicletniolog-y-of--slwJx..L..ot._those-fl'onts-fV~an-den-Boseh-et-a/•.,
....
1988a,b, 1989). However, because these
analytical models assume constant dispersal
properties of disease and constant spore
production rate and infectivity they are not very
practical. For more practical types of analysis in
spatial epidemiology, simulation approaches arc
needed. For instance, Zawolek ( 1989) uses
simulation to investigate the consequences of
·--------'A·----·---------·-------·----········; :a··············----··························:
I :
stochasticity and dual
spore
dispersal
mechanisms on formation of plant disease foci.
Earlier on, enlightening simulation studies with a
spatial nHxlel were conducted by Zadoks and
These
simulation
Kampmeijer
( 1977).
approaches which were developed at the field
level, were in a very preliminary sense also
applied at the (supra-)regional level.
rc·······················-···················: :·a··············-··············--·············
1
i!!;;ii:n::::t
1
::!!!i!lli~lllllii!!H..
;:!!!!:",
:·e··-·--·············i······---·········:;·:·r
,
'!, 1
;!,'
1 11
.,,
. i;>·,
::·:
~
I
'
' :::::::::::::.:::::::::::::::::::::::::::::::::::: ::::::::::::::::: ::::·:::.: :::::::::.:::::::::::::: ················-···--·-·······················;
·····-···~······4·-~---···················· ···- ·············-·································;
··-··-·-··········-!..... !.. ~ .....................!
:::::::::::::::::::::::::::::::::::::::::::::::1
Ill
I
!
!
:
l
'LnH::~,
,'
..::::::!li:i;;;::; 1;:,:...
',Udn:;:;!:;::nu'tt", '
'
. ... It:.·.::·.:: . . :.;·.
t
tl
:
I
' :
:,,
I
llt
I
··:,;:,·.:.:.· .·.:r~.·:·.:".·:· :"
'"
!
''i
1
j
·:::::::;!1~l1!~li!i!:·::· ' :" .,',,·~,' :,::,;';'.' ,::: ' ! :·::·:.};: :::,: ::'':::::: ·.!: I
:.............................................J :. . -.. . . . . . . . . . . . . . . ·-.· · · · : :. . . . . . . . . . . . . . . . . . . . . . . . .::......................... ~ . .··········.......:...........~ ............:~ .:...................!
I
Figure 9.
', ' ';:"'
H
'I
Simulated spatial spread of beet yellows virus in sugar beet by dispersing vector aphids,
Myzus persicae. The simulation shown here represents an experiment in which three plants in the
centre of a 12 x 12 m2 field plot were inoculated with BY'/ on 23 June and after two days infested
with 9 M. persicae. Duplicate gmphs indicated by the letters A, B, C ... show the simulated state of a
plot at two week intervals. The upper graph of each 'duo' represents the spatial distribution of
virus-infected plants, while the lower gmph gives the spatial distribution of the aphids. In the upper
figure, numbers indicate the proportion of leaf area infected with virus: (0): 0-0.1; ( 1): O.l-0.2; etc. A
shaded block indicates that the plant has symptoms. The type of shading indicates t11e proportion of
infected leaf area: ( ;;;>: 0-0.3; ( ~): 0.3-0.6; ~1: P.6-0.9; ( . ): >0.9. In the lower figure, numbers
indicate the density of aphids Qer plant. Densities higher than 9 per plant arc indicated with a shaded
block (~g).
~-------
-~~--~~~-~--~-~~~~~
SYSTEMS MANAGEMENT
Farm management requires a major
managerial effort at various levels of integration.
We distinguish the levels of the farmer and the
policy maker. At both levels the overall
objectives are to ensure economically and
ecologically sustainable agricultural production
in 01e long term.
At the farm level a farmer's decisions
include choice of crops to be grown, timing of
selling, buying or storing products, and
long-term
investments
in
buildings,
infrastructural changes and fanning machinery.
At this level, decisions are greatly affected by
market situation and commodity prices.
Ecological considerations have to be taken into
account to guarantee the sustainability of the
farm and to protect the environment. Constraints
at 01is level comprise public regulations and
private considerations on what is ecologically
acceptable and sound. Decisions at the farm
level set the constraints for decisions at the crop
level.
At the crop level, OlC farmer must decide
on crop variety, and on the timing and methods
of seed-bed preparation, planting and harvesting,
amount and frequency of fertiliser application
and control of pests, diseases and weeds. The
majority of these decisions arc based on
experience, perception and preference, and can
be supplemented by insight supplied by
agricultural research, education and extension.
At the policy level, various instruments are
available for stimulating desired developments in
crop protection at the (supra-)regional level.
Product price regulations, land set-aside
schemes, subsidies on inputs, and cultivar or
pesticide registration procedures are but a few
examples. Although discussions tend to focus on
the instruments, first decisions on the objectives
of crop protection policy as part of agricultural
policy should be made. General objectives
concerning sustainability should be made
operational to enable evaluation of the
effectiveness of the various instruments.
The current availability and usc of external
resources in agricultural production, such as
artificial fertilisers, pesticides and machinc1y, is
unprecedented in human history. As a
consequence, historical experience is of little use
for optimising
management of current
agricultural production systems. Since the 1950s
quanlilalivc knowledge of biological and
agronomical consequences of a farmer's
decisions increased dramatically, not only in
industrialised agriculture but also in low input
agriculture in many developing countries. An
important part of these decisions concern crop
protection. Pesticides have been the main
instrument in crop protection since 01eir large
scale introduction, almost half a century ago.
Decisions to apply pesticides usually occur at the
end of a chain of decisions leading to a certain
crop in a certain field. No feedback exists with
decisions on, for instance, nitrogen fertiliser
rates which may affect 01e occurrence of
economically damaging levels of growth
reducing factors. Thus, chemical pest control
became the 'fudge factor' of crop production, and
was used to correct previous agronomic decision
en·ors. 1l1is role was made possible by the fact
that costs of chemical control usually concern a
small fraction of total production costs for a
farmer. However, the use of pesticides causes
major public costs, such as poisoning of
non-tat·get organisms, including man, pollution
of soil, air and water, and occurrence of
pesticide-resistat1t noxious organisms. Public
concern about these issues has resulted in
reconsideration of the objectives of crop
protection, and the place of crop protection
decisions in the hierarchy of agronomic
decisions at field and farm levels. Increasingly,
crop protection is considered as an integrated
part of on-farm decision making which aims at
optimising both economic atld ecological
sustainability criteria. To assist in improving
decision making in crop protection, production
ecological approaches have been developed.
Decision support at the farm level
Decision support at the farm level has been
oriented predominantly towards solving tactical
decision problems. Recently, tl1e much broader
goal of improving the general level of production
ecological knowledge Ou·ough training has
allracted attcallioll.
"-~----~~"-~---""-~ ----~~~---~----~
Production
support
ecology and
tactical decision
damage at lower attainable levels of crop
production. Similar studies have been canied out
Sound systems of tactical decision support
for other growth reducing factors, e.g. weeds
comprise four building blocks: epidemiological
(Kropff and VanLaar, 1993).
relations, damage relations, sampling und
Key element in any supervised control
monitoring
methods and cost/benefit/risk
system is an appropriate sampling and
monitoring procedure to quantify the density of
analyses. An example of such system is the
supervised control system EPIPRE, which
the damage causing organism. Sampling and
provided recommendations on aphid and disease
monitoring approaches pragmatically utilise
sample size and sampling frequency to
control in winter wheat in the Netherlands for
over 15 years (Zadoks, 1989). We will usc this
compensate for the lack of understanding or
system to illustrate the role of production
predictability of system dynamics. Statistical
ecology in farmer oriented decision support.
theory and biological expertise are combined to
To develop epidemiological relations for
derive sampling protocols which meet the
forecasting, stage-structured simulation models
requirements of simplicity, biological soundness,
were first used to investigate the extent to which
low labour requirement, compatibility with other
the population dynamics of aphids and diseases
crop management activities, and low costs
were understood. For example, quantitative
relative to the value of the product (Rabbinge
and Mantel, 1981). These requirements also
models of the diseases Septoria tritici, Erysiphe
graminis, and Puccinia recondita showed that
apply to the monitoring protocol, the sequence of
far fewer ·lesions developed in field experiments
sampling protocols in the course of a growing
than was expected, using life-table data (Internal
season. In EPIPRE, forecasts of pest density and
Reports, Department Theoretical Production
associated damage were updated using the
Ecology). Moreover, the difference between
sampling information on pest incidence collected
simulated and observed disease progress
by the farmer. In this way, field-specific
prediction errors were kept in check. This
appeared to vary per field. Sensitivity analyses
of the epidemiological models showed U1at field
updating procedure is known in operations
specific factors affecting spore production, spore
research as decision-making with rolling
dispersal and host penetration were insufficiently
planning horizon. Sample size, the number of
understood or not quantified accurately enough.
tillers or leaves to be inspected, was chosen such
Therefore, for forecasts in EPIPRE, simple
that an acceptable coefficient of variation would
exponential models were used which were
be attained at densities close to U1e damage
ttu·eshold. To minimise sampling effort
calibrated on field data using regression analysis.
absence/presence sampling methods were
Damage mechanisms of wheat diseases and
aphids were studied in laboratory and field
developed. Sampling intervals were adjusted to
the actual situation so as to account for crop age,
experiments, using mechanistic models of crop
disease severity and its rate of change. Recently,
growth as a means of quantitative hypothesis
furU1er advances have been made in minimising
testing. For Puccinia recondita and Erysiphe
graminis these studies served to corroborate and
not only the size of one sample, but also the
extrapolate empirical damage relations used in
number of sampling occasions needed during a
forecasting. For aphids, damage relations derived
growing season (Nyrop and Van der Werf,
1993). Basically, the smaller the estimated pest
from a wheat growth model into which uptake of
assimilates by aphids and decreased leaf
intensity during a sampling occasion relative to
the damage threshold, the longer the interval
photosynthesis by honeydew were introduced,
before a next sampling need be taken. More
were as accurate as empirical damage relations
uncertainty about future progress of the
derived from 15 years of field data obtained on
more than 20 locations (Rossing, 1991 ). The
population and the concomitant damage results
-~··--------·-----mechanistic__nature of the_model and t.he__ln__sbm1el:___saffipling intervals. The authors
inclusion o( the effect of nitr()geir short:Jge onshowed - that such 'multipa.rtite sequential
----- -~······ ---growth-and-development-;-enablcd-assessments-of~--classificatiolt'------or--__~.adapti-ve-----frequeney
38
classification' can decrease the risk of erroneous
(in)action which is caused by uncertainty in
sampling estimates (Figure 10).
Analyses of costs, benefits and risks are
needed to select the decision which best meets
the objectives of the farmer. Recommendations
to reflect
are
often
implicitly
tuned
risk-averseness of the decision maker, but no
explicit information is supplied on the
uncertainty of the forecasts. An alternative
approach is to present the riskiness of decision
alternatives (Figure 11), and leave the choice of
the best decision to the farmer (Rossing et al.,
1994).
Production ecology and training at tile farm
le~·el
Evaluations of the EPIPRE system (e.g.
Blokker, 1983) showed that farmers appreciated
density
0 .• high
intervene directly
the educational aspects of EPIPRE even more
than the (marginal) financial savings. Learning
about the crop and its relations with the
environment appears prominently in modern
systems of farmer-oriented decision support. For
instance, successful introduction of IPM in
Indonesia was achieved through field schools of
farmers at the village level, in which farmers
exchanged experiences on level and regulation
of pests and their natural enemies, and
consequences for yield (Van de Fliert, 1993).
Somers and Rtlling ( 1993) evaluated a number of
projects in the Netherlands aimed at reducing the
level of nutrient and pesticide inputs per unit
area. They conclude that the adoption of such
innovative production systems can only be
achieved through a learning process which
involves both farmers and extension agents, and
during which there is a gradual change of
B
tilne
A
2
0
Samples
Figure 10.
0
0
0
Time
Example of a pest monitoring method that schedules sampling in time in a flexible
way. Sampling is done sequentially and uses the double SPRT method (Figure lOA). (SPRT is the
acronym for Sequential Probability Ratio test.) The double SPRT method has three alternative
decisions: (0): density is currently high and requires an intervention; (2): density does not require
immediate intervention, but sampling should be repeated soon to make sure that pest density still falls
within acceptable limits; (3) pest density is quite low such that rcsampling can be postponed further
into the future. Depending upon the outcome of sampling, the pest population will be monitored more
or less frequently (Figure lOB). The performance of monitoring schemes using double SPRT or
alternative sampling protocols can be simulated. The performance criteria are: 1. probability of
intervention; 2. number of sampling bouts actually scheduled; 3. total number of samples in all bouts;
---"4'*-',-tCH-mHlati¥e-pcst/-di.~ease-dcnsit7'-(a-v-er:ag~<.Lpcrccutilcs)_;_A densi I ¥-JlUn tervention (average and
~~rcentil~~)~~The QUIC<!_Illes _of l)uch~~il!lJ!!~j()ns arc similar to field tests of the monitorin~ protocol
~-~applie1hu-Europl~nrrTed-mite--;-Panvnydms-u/ml;-intipplcs:- ~- ~ -- -~ - ~-~ ~ ~ -- ~~ - ~~ ~~ ~~ ~ ~ ~ - ~-
Frequency
A: No Control
t200
E01500
B: Immediate Control
400
300
200
100
Wd
~Ynn
0
0
200
E0-1-
~
400
600
000
tooo
t 200
Costs (011/ha)
Figure ll. Simulated frequency distribution of
costs of no chemical control (A) and immediate
chemical control of aphids and brown rust jointly
(B) in winter wheat, in 500 Monte Carlo runs.
Initial state of the system: crop development
stage 'full ear emergence', aphid incidence 5% of
tillers, brown rust incidence 2% of leaves.
Expected costs in (A) and (B) (indicated by
arrows) are the same, implying that the initial
state of the system represents a damage
threshold.
objectives, targets and tisk perception. This
process differs frorri the adoption of single
production techniques, such a'i mechanical
weeding, in which the rate of adoption depends
on the technical and entrepreneurial inclination
of the farm community. Optimal adjustment of
crop protection decisions in relation to other
farm management decisions, represents a
· complex decision problem which requires highly
skilled and trained farmers. As a consequence
the role of extension changes from, traditionally,
technology transfer and problem solving to
facilitation and human development (Somers and
Roling, 1993). In the examples from Indonesia
learning process, raU1er than to provide
prescriptive recommendations. Thus there is a
shift from top-down to bottom-up approaches.
Production ecological knowledge can play
an important role in these learning processes,
since qmmtitative models provide a fmmework
for evaluating 'what-if questions ·about
consequences of different levels of pest attack
during various periods of the growing season.
Well-tested models can be used to explore crop
response
under
various
environmental
conditions. Such exploration helps to develop the
intuition of decision makers with regard to
ecological relations in a crop system. Similar
applications of production ecological principles
have been developed for formal teaching at the
Wageningen Agricultural University (Rabbinge
and Eveleens, 1993) and elsewhere. The
exphmatory nature of U1e models represents an
essential requirement since it enables application
under widely different environmental conditions.
Development of a suitable interface between
model and trainee calls for both production
ecological knowledge and agricultural education
expertise, and usually requires considerable
intellectual and progmmming l.abour.
Decision support nt the policy level
In U1e area of policy~oricnted decision
suppott explorative land use studies have shown
in
demonstrating
the
much
potential
consequences for land use of explicit objectives
with respect to economic and ecological
sustainability. Because ecological and short -term
economic objectives are often conflicting,
formulation of agricultural land use policy
involves searching for acceptable compromises.
Then, questions arise concerning the trade-off
between, for instance, minimising pesticide input
and maximising production levels in different
production situations. To answer these questions,
quantitative knowledge on production technical
and production ecological relations in production
systems is needed, in combination with
information on Ute suitability of an area for
different types of land use. The suitability of an
area is assessed in a qualitative land evaluation.
During this phase, infmmation on climate, soil
···-···--·-·-----a-nd-theN·ettrerhnrds~olc-of-thc-extcnsiun-----fcatttt'Cs-sueh-as-tex1ttre;-sloiJe-a11d-tiFainage,alld~-
- -service- was -to-part icipatc -in -and--facili talc-the-
infrastructure is combined to arrive at . an
----------------------~------~~~~~~----------------
-- ~~-~-----~~~------· - --
assessment of the production situation. In the
next, quantitative phase, the production situation
combined
with
predefined
production
technology results in a production level. For
each production situation, production levels
under irrigated, rain-fed and nutrient limited
conditions can be distinguished. Product.ion
levels arc formulated according to the concept of
'best technical mc~ms', implying that inputs are
employed as efficiently as possible. For the
definition of production technologies primary
and secondary inputs are distinguished. For
primary inputs such as water and nutrienl<;
substitution is scarcely possible. For secondary
inputs, such as weeding technologies,
substitution is possible. Depending on the time
horizon of the study, current, experimental and
theoretical production technologies m·e included.
Finally, a linear programming model is used to
find the mixture of production systems which
provides the optimal compromise of objectives.
In a study of options for land use in the
European Community the Netherlands Scientific
Council for Government Policy (1992) analysed
the scope for agricultural policy in the EC in the
next 20 to 25 years. Four policy scenm·ios were
formulated, based on main positions in the
cunent political debate: (1) free trade m1d free
market, (2) regional development, (3) nature and
lm1dscapc and (4) environmental protection. The
outcomes of the four scenarios showed that the
continuing rise in productivity results in
increasing land surpluses and further loss of jobs
in agriculture, irrespective of policy. It also
appeared that there exist good possibilities for
more environmentally friendly agricultural
production, which uses less than 25% of the
current volume of active ingredients and
fertiliser nitrogen. The concomitant production
systems are characterised as highly productive
with maximum use of ceo-technological
principlcs and biological purgatory tools, such as
resistant and tolerant cultivars, high recovery of
nutrients, biological control, sound crop
rotations, 'catch crops' to minimise nutrient
leaching, and combination of m·ablc and animal
husbandry with approximately closed nutrient
cycles. Supra-regional hmd-use studies arc
~ c•rvtsrrgc-o-rurolhcr-rcgi<>ns~---c.g-:-Tmml-r'-maum-
-~~~~~-~--
Land evaluation studies, whether at the
supra-regional or at the regional and farm levels,
have a bright future. They provide a framework
for exploring management options at the farm
and higher levels of integration, utilising
production ecological knowledge at the crop and
cropping system level. At the same lime they
provide an instrument for research prioritisation,
by determining the relative importance of the
lacks of information about various hspects of
production systems for the outcome of the study.
In conclusion, the production ecological
approach provides instruments for systems
studies at different levels, which can be used to
clarify the consequences of decision alternatives
for the objectives of management, and thus
provide a rational basis for decision making.
INSTITUTIONALISATION OF
PRODUCTION ECOLOGICAL
APPROACHES
The production ecological approach
requires a new type of agronomists, and more
specifically, also a new breed of crop protection
scientists. Systems analysis comprises an
analysis phase m1d a synthesis phase. The
analysis phase is compatible with the traditional
reductionist scientific method, which produced
the disciplines entomology, phytopathology,
nematology and virology. The synthesis phase,
however, necessitates a gcneralistic approach
with interdisciplinary work at different
integration levels.
This means that the
traditional borders between, for example,
entomology and phytopathology disappear, and
organism-oriented approaches need to be
replaced by system-oriented methods. The result
may be crop protectionists who combine
sufficient understanding of specific pathosystems wilh affinity for general unifying
concepts and their application in practical crop
protection. An example of capacity building in
production ecology along t11ese lines is the
ongoing project SARP, short for Systems
Analysis and simulation in Rice Production. We
will discuss the approach in this project and
-----<>ullli1c-nsllii<rTc-rlnresulfs:--
-India. -
~ ----------~------
--------- ~--
.
'tl
Outli~e
of the SARP project
SARP was started in 1984 by national
agricultural research centers (NARCs) in south,
east and south-east Asia, the Centre for
Agrobio1ogica1 Research in Wageningen (now:
Research Institute for Agrobiology and Soil
Fertility (AB-DLO)) and the Department of
Theoretical Production Ecology of the
University
Wageningen
Agricultural
(WAU-TPE), in collaboration with the
International Rice Research Institute (IRRI).
Aim of the project was and is to build research
capacity in the field of systems approaches at the
NARCs and at IRRI with the help of modern
systems research techniques. This goal is rooted
in the proposition that research efficacy and
efficiency can be much enhanced by a systems
approach. The long-tenn goal is to further
enhance sustainable productivity of rice-based
systems. Staff time is contributed by
participating institutes. Funds for training,
exchange of scientists, and coordination m·e
contributed by the Directorate General for
International Cooperation (DGIS) of the Dutch
Ministry of Foreign Affairs.
During the first two phases, before 1991,
three training programs were held under SARP
auspices. In total 91 researchers from 16 NARCs
throughout south-east Asia and IRRI (Figure 12)
were trained in the usc of systems analysis and
computer simulation modelling as a tool in their
research activities. Some of these teams have
later organised their own national training
courses for sister institutes, the so-called second
generation teams (Figure 12). Training was
followed by case studies within the informal
SARP network, to actively introduce the
approach in the NARC's research programs.
Case study topics were selected by the
participants in accordance with ongoing resem-cl1
at their institutes. During the case studies the
teams were visited by SARP staff for technical
and scientific support.
Trainees were always part of a team with a
critical mass of at least four researchers from the
same NARC. At least one member was a crop
physiologist. The other members came from
various disciplines. Each tcmn had the support of
policies. One of the scientists in the team was
selected to be team leader. Each team consisted
of different disciplines to ensure a focus on the
rice system, mther than carrying out research in
the traditional disciplines, with a major risk of
neglecting impot1ant system aspects. Various
workshops and a closing conference in Bangkok
were organised. A substantial number of
publications show the results of the training
period 1984-1991. A detailed review of the
project is given by Ten Berge (1993).
SARP's third phase covers 1992 to 1995.
Emphasis is placed on collaborative research.
From the training programs four themes emerged
as a framework for collaborative research:
•
•
•
•
agro-ecosystems: agro-ecological zonation,
timing of crops m1d crop sequences,
optimisation of regional water usc;
potential production: crop responses to
light and temperature, development m1d
morphogenesis;
crop and soil management: water and
nitrogen mm1agemcnt for different soils,
rice varieties, plant spacing, plant
establishment;
crop protection: mechahisms of dmnage by
pests and diseases.
The aims in the third phase of SARP are:
•
•
•
•
to reinforce the teams and to support joint
research progrmns within the informal
network;
to develop applications at the crop and
agro-ecosystem level aimed at both policy
on
makers
(e.g.
through
studies
agro-ccological zonation), and extension
workers m1d farmers (e.g. by directing
research to development of tools for advice
on nitrogen and pest and disease
management);
to support national training programs when
they m·ise;
to transfer coordination of the project to
NARCs and IRRI.
Participants in the crop protection theme
prioritiscd a few pathosystcms for joint research:
~ ~-~~~-a-senTor~-sctcnt[~-~thc -fcam~~~upcrviSor~wllo~~- -~-ilicTii:Sccrp-iiffi()sysic-m1Ice-:=stemooier(S~B~;Tive
~~-mainlainedTlosc
ltnks-w ittr-th-e-NJ\RC's -rescarc h
~-spec i~cs--m·c -~~of ·u-cconom ic~u importance).~ -~the
0
t
t
t
t
1
It
t
t
t
It
t
t
t
I
I
It
f
I
I
I
I
Itt
It
I
I
I
I
o
I
I
I
I
I
I
It
I
I
I
I
It
I
I
I
I
I
I
It
I
I
I
I
I
I
I
0
I
I
I
I
I
I
I
I
It
I
I
SARP teams
•
0
first generation
second generation
per November 1991
Figure 12. Geographic distribution of National Agricultural Research Centres (NARCs) that
pru-ticipated in training programs of SARP: Systems Analysis and simulation in Rice Production.
optimally utilise the project's network character
bacterial pathosystem rice-bacterial leaf blight
(Rossing et al., 1993).
(BLB, Xanthomonas campestris pv. oryzae), aild
the fungal pathosystem rice-sheath blight (ShBI,
Rhizoctonia so/ani). Criteria used in the selection
SAI~P's impact on NARCs
The mid-tetm results of SARP's approach
procedure included the number of teams that was
for institution building were evaluated in a
already actively involved in resemch on the
pathosystem (reflecting, among others, the
workshop organised in December 1993 by
economic importance of the pest and the
ISNAR under co-sponsorship of ICASA
disease), the current state of knowledge on
(International Consortium for the Application of
Systems approaches in Agriculture) and the
damage mechanisms, and the scientific support
Dutch Ministry of Development Cooperation
available at IRRI and in Wageningen. A research
agenda was developed during workshops and a
(Goldsworthy and Penning de Vries, 1994). The
institutionalisation process m1d the collaboration
priority was given to research into dmnage
between NARCs, IRRI, WAU-TPE and
mechanisms and field experiments aimed at
t<~sting the crop models. Since quantification of
AB-DLO was characterised by ten features, that
damage mechanisms often requires specialised
contributed substantially to the successful
equipment, appropriately equipped institutes
assimilation of systems approaches by the
_____wereJdentified. For Jhc vaJJgation
_9_~~ri_met!~~ _____t-.f_~~~~----------- ___________
was agreed
upon to
- ---- - - ----------- ----
..:
•
training of teams, rather than individuals
commitment of the NARCs to integrate
systems analysis in their institutional
approach
intensive initial tJ·aining during six weeks
case studies by teams in their home
institutes
backstopping of the teams
guaranteed availability of appropriate
equipment and funding for experiments
communication between NARCs, lARCs
and advanced research organisations
networking initialised by IRRI
continuous involvement of the supervisors,
representing the management of the
institutes; and last but not least
a very enUmsiastic group of scientists
convinced of the need for a systems
approach and capable of creating a
productive atmosphere.
Goldsworthy, P.R. and Penning de Vries, F.W.T.
(1994). Proceedings of the /SNAR!ICASA
Workshop November 22-24, 1993, The Hague.
Kluwer, Dordrechl. (In press).
Groenendijk, C.A .• Van dcr Werf, W., Van Dijk, E.
and Carneiro, R.A. (1990). Mededelingen
Faculteit
der
Landbouwwetenschappen
Rijksuniversiteit Gent, 55: 1085-1098.
Hurcj, M. and Van der Wcrf, W. (1993). Annals of
Applil•d Biology, 122: 201-214.
Janssen, A. and Sabclis, M.W. (1992). ExperimenJal
and Applied Acarology, 14: 233-250.
Jcttcn, Th. H. and Takkcn, W. (1994). Change,
18: l0-12.
Kropff, M.J. and Van Laar, H.H. (1993). Modelling
Crop- Weed Interactions. CAB-International,
Wallingford, and International Rice Research
Institute, Manila. 274pp.
Kropff, M.J. Vossen, F.J.H., Spillers, C.J.T. and
Groot, W. de (1984). Netherlands Journal of
Agricultural Science, 32: 324-327.
Mols, P.J.M. (1993). Ph.D. Thesis, Wagcningen
The production ecological procedures and
Agricul!ural University, Wageningen. 203 pp.
:approaches that were implemented in many
Monteith. J.L. (1977). Philosophical Transactions
research centres in Asia will help to introduce
Royal Society London, Series B, 281: 277-294.
economically competitive, environmentally
Netherlands Scientific Council for Government
friendly,
agronomically
advanced
and
Policy. (1992). Ground for Choices. Sdu
ecologically sound integrated pest and disease
Publishers, The Hague. 144pp.
management systems. The ultimate aims of
Nyrop, J.P. and Vander Wcrf, W. (1993). /landbook
agricultural research may thus he brought one or
of Sampling Methods for Arthropods Pests in
two steps nem·er.
Agrimlture. (Pedigo, L.P and Buntin, G.D.,
eds.). CRC Press, Boca Raton, Florida.
pp 245-336.
REFERENCES
Penning de Vries, F.W.T. and Van Keulen, H. (1982).
La productivite de paturages Sahelians (Penning
Bastiaans, L. (1991). Phytopathology, 81: 611-615.
de Vries, F.W.T. and Djiteye, M.A., eds.).
Bastiaans, L. and Kropff, M.J. (1991). Systems
Agricultural Research Reports 918. Pudoc,
Simulation at IRRI. (Penning de Vries, F.W.T. et
Wageningcn.
a/., eds.) IRRI Research Paper Series Number
Rabbingc, R. (1976). Biological Control of Fruit-tree
151. pp 21-26.
Red Spider Mite. Simulation Monograph. Pudoc,
Bastiaans L. and Teng, P.S. (1994). Submitted.
Wagcningen. 228pp.
Blokkcr, K.J. (1983). Ph.D. Thesis, Wagcningen
Rubbingc, R. (1986).
Netherlands Journal of
Agricultural University. 389pp.
Agricultural Science, 34: 239-251.
Rabbingc, R. (1993).
Crop Protection and
Conway, G.R. (1984). Pest and Pathogen Control:
Strategic, Tactical and Policy Models. IIASA,
Sustainable Agriculture (Chadwick, D.J. and
Laxenburg, Austria.
Marsh, J., eds.). Wiley, Chichester (Ciba
Foundation Symposium 177). pp 2-29.
Fransz, H. G. ( 1974). The Functional Response to Prey
Density in an Acarine System. Simulation
Rabbinge, R. and De Wit.. C.T. (1989). Simulation and
_______ Mo!lQgrll}2h_.J~!!Q_Q£,~_ng~tiu_g~nj42J)Ik____________ _________ s~vsJem:c __ ManagemeuL-irL---Cr.op___£wlecJion._ --
•
•
•
•
•
•
•
•
\
44
--------------------1
(Rnhbinge, R., Ward. S.A. and VanLaar, H.H.,
eds.). Pudoc, Wugcningen. pp 3-15.
Rabbingc, R. and Evelcens, K.G. (1993). European
Journal of Agricultural Edm:ation and
Extension (In press).
Rabbinge, R. and Mantel, W.P. (1981 ). N£•therlands
Journal of Plant Pathology, 87: 25-29.
Rabbinge, R. and Rijsdijk, F.H. (1981). Effects of
Disease on the Physiology of the Growing Plant
(Ayres, P.G., ed.). Cambridge University Press,
Cambridge. pp 201-220.
Rabbinge, R., Ward, S.A. and VanLaar, H.H. (1989).
Simulation and Systems Mmwf(ement in Crop
Protection. Pudoc, Wagcningcn. 420pp.
Rossing, W.A.H. (1991). Netherlands Journal of Plant
Pathology, 97: 87-103.
Rossing, W.A.H., Danmcn, R.A. and Hendrix, E.M.T.
(1994). Crop Protection, 13: 2~-34.
Rossing, W.A.H., Rubia, E.G., Hcong, K.L.
Keernti-Kasikorn, M. and Reddy, P.R. (1993). .
SARP Research Proceedings. 20lpp.
Rubia, E.G. and Penning de Vries, F.W.T. (1990).
Journal of Plant Protection in the Tropics,
7: 87-102.
Sabelis, M.W. (1981). Biological Control of
Two-spotted Spider Mites Using Phytoseiid
Predators. Part/. Agricultural Research Reports
910. Pudoc, Wageningcn. 242pp.
Schnns, J. (1992).
Ph.D. Thesis, Wagcningen
Agricultural University. 115pp.
Secher, B.J.M., Rossi, V. and Battilnni, P. (1993).
Proceedings
of
the
Workshop
on
Computer-based DSS in Crop Protection.
Parma, Italy 23-26 November 1993. Sp-Rcport,
Danish Institute of Plant and Soil Science No.7.
223pp.
Somers, B.M. and R<>ling, N.G. (1993). Knowledge
Devl'lopment for Sustainable Agriculture (in
Dutch). NRLO-Report 93/12. 72 pp.
Spedding, C.R.W. (1990). Theoretical Production
Ecology. (Rabbinge, R., Goudriaan, J., Van
Keulen, H., Penning de Vries, F.W.T., and Van
Laar, H.H., eds.). Pudoc, Wageningen.
pp 239-248.
Spitters, C.J.T. (1984). COLUMA (Comite Francais
de Lulie Contre les Mauvais llerbers)-EWRS
(European Weed Research Society), Paris.
pp 13-27.
Spitters, C.J.T, van Keulen, H and van Kraalingen,
D.W.G. (1989). Simulation and Systems
Management in Crop Protection. (Rabbinge, R.,
Ward, S.A. and Van Laar, H.H., eds.). Pudoc,
Wageningen. pp 147-181.
Sutherst, R.W. (1991). Review of Agricultural
Entomology, 79: 1177-1187.
Ten Berge, H.F.M. (1993). Systems Approaches for
Agricultural Development (Penning de Vries, F.,
Teng, P. and Metselaar, K., eds.). Kluwer,
Dordrecht. pp 515-538.
Van de Fliert, E. (1993)./ntegrated Pest Management:
Farmer Field Schools Generate Sustainable
Practices. Wageningen Agricultural University
Papers 93-3. 304pp.
Van den Bosch, F., Zadoks J.C. and Metz, J.A.J.
(1988a). Phytopathology, 78: 54-58.
Van den Bosch, F., Zadoks J.C. and Metz, J.A.J.
(1988b). Phytopathology, 78: 59-64.
Van den Bosch, F., Metz, J.A.J., Prinking, H.D. and
Zadoks, J.C. (1989). Phytopathology, 78: 59-64.
Van Keulen, H. and Wolf, J. (1986). Modelling of
Agricullural Production: Weather, Soils and
Crops. Pudoc, Wageningen.
Van Oijen, M. (1991). Ph.D. Dissertation.
Wageningel) Agricultural University.
Van Roermund, H.W.J. and Spitters, C.J.T. (1990).
Netherlands Journal of Plant Patlwlogy,
96: 17-28.
Van Roermund, H.W.J. and Van Lenteren, J.C.
(1994). The functional response of the whitefly
parasitoid Encarsia formosa on the plant level.
Proceedings of the Experimental and Applied
Entonwlogy, N.E.V. Amsterdam, Vol. 5, 1994.
(In press).
Van dcr Werf, W., Bastiaans, L., Rossing, W.A.H. and
Rabbinge, R. (1990). Proceedings of the
International Symposium on Biotic Stress of
Barley in Arid and Semi-arid Environments.
Montana State University and !CARDA.
pp 87-102.
Van der Werf, W., De Moed, G.H. and Smits, P.H.
(1991)./0BCIWPRS Bulletin, XIV(7): 108-115.
Van der Werf, W., Nyrop, J.P. and Hardman, J.M.
(1994). Aspects of Applied Biology, 37. (In
press).
VanderWerf, W., Rossing, W.A.H., Rabbinge, R., de
Jong, M.D., and Mols, P.J.M. (1989).
--SpiUcrs,-C.J.~--(-l-900}.---AGta-llot'tiGulturae-,--------PARASI-1~/S-88-({;avalloro,R.-and-DeluCGhi,V.,
______________________1_67: 342-368.
edsj_p_p_8_2:_119.
Zadoks, J.C. (1989). Plant Disease EpidcmiolORY·
Volume II (Leonard, K.J. ruu.l Fry, W.E., eds.),
pp 3-29.
Zadoks, J.C. (1993). Crop Protection and Sustainable
Agriculture (Chndwick, D.J. and Marsh, J.,
eds.). Wiley, Chichester (Ciba Foundation
Symposium 177), pp 48-60.
Zadoks, J.C. and Kampmcijer, P. (1977). Annals of
the
Nt•w
York
Academy
of Science,
287: 164-190.
Zadoks, J.C. nnd Schein, R.D. (1979). Epidemiology
and Plant Disease Management. Oxford
University Press, New York. 427pp.
Zawolek, M.W. (1989). Ph.D. Thesis, Agricultural
University, Wagcningen. 230pp.
46